The physical GPU-cloud atlas
Neocloud locations
Neocloud Map pins 126 GPU-cloud sites across 26 countries and 17 operators, from CoreWeave, Nebius, Lambda and Nscale to Crusoe, IREN and Vultr. 84 are operating today. Filter by city, delivered and planned MW and GPU platform, then open any site for its dated first-party source. Method & sources
Compare physical locations
Delivered MW is live or billing capacity; planned MW is the contracted or announced total. Each site page states the exact power scope and its source. Use Columns for GPU counts, cooling and evidence dates.
| Applied Digital Delta Forge 1 | Applied Digital | Boyce | United States | Infrastructure developer | Under construction | — | 3001 | Contracted critical IT load; hyperscaler unnamed. | — | 2026-11-11 |
|---|---|---|---|---|---|---|---|---|---|---|
| Applied Digital Polaris Forge 2 | Applied Digital | Harwood | United States | Infrastructure developer | Under construction | — | 2001 | Contracted critical IT load; hyperscaler unnamed. | — | 2026-11-11 |
| Core42 Dallas | Core42 | Dallas | United States | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| Core42 Lake Mariner | Core42 | Barker | United States | GPU / AI deployment | Operational | 601 | — | Core42 production AI infrastructure after expansion; separate from Fluidstack leases. | AMD Instinct MI300; additional NVIDIA and AMD infrastructure | 2026-11-11 |
| Core42 Minneapolis | Core42 | Minneapolis | United States | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| Core42 Stockton | Core42 | Stockton | United States | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| Core42 Sunnyvale | Core42 | Sunnyvale | United States | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| CoreWeave Arasur | CoreWeave | Arasur | Spain | GPU / AI deployment | Announced | — | 181 | Reported future lease allocation, not confirmed operating MW. | — | 2026-11-11 |
| CoreWeave at Bulk N01 | CoreWeave | Vennesla | Norway | GPU / AI deployment | Announced | — | — | — | NVIDIA GB200 NVL72 | 2026-11-11 |
| CoreWeave at Denton | CoreWeave | Denton | United States | GPU / AI deployment | Operational | 2603 | 2603 | About 260 MW billing of 260 MW leased critical IT (Core Scientific, 5 Nov 2026). | — | 2026-11-11 |
| CoreWeave Austin | CoreWeave | Austin | United States | GPU / AI deployment | Operational | 201 | 201 | About 20 MW billing of 20 MW leased critical IT (Core Scientific, 5 Nov 2026). | — | 2026-11-11 |
| CoreWeave Barcelona Zona Franca | CoreWeave | Barcelona | Spain | GPU / AI deployment | Operational | 151 | — | 15 MW facility capacity at the May 2025 inauguration. | NVIDIA H200 | 2026-11-11 |
| CoreWeave Copenhagen DEN01 | CoreWeave | Copenhagen | Denmark | GPU / AI deployment | Operational | — | 22.51 | Planned IT capacity; majority online, exact delivered MW undisclosed. | NVIDIA Blackwell | 2026-11-11 |
| CoreWeave Crawley | CoreWeave | Crawley | United Kingdom | GPU / AI deployment | Operational | — | — | — | NVIDIA H200 | 2026-11-11 |
| CoreWeave Dalton | CoreWeave | Dalton | United States | GPU / AI deployment | Operational | 751 | 1751 | About 75 MW billing of 175 MW leased critical IT (Core Scientific, 5 Nov 2026). | — | 2026-11-11 |
| CoreWeave Douglasville | CoreWeave | Douglasville | United States | GPU / AI deployment | Announced | — | 91 | Announced allocation at each facility in the 2023 agreement. | — | 2026-11-11 |
| CoreWeave Ellendale Polaris Forge 1 | CoreWeave | Ellendale | United States | GPU / AI deployment | Operational | 1751 | 4001 | Contracted critical IT load across the campus; phased delivery. | — | 2026-11-11 |
| CoreWeave Falun | CoreWeave | Falun | Sweden | GPU / AI deployment | Announced | — | — | — | NVIDIA Blackwell (announced cluster) | 2026-11-11 |
| CoreWeave Helios | CoreWeave | Afton | United States | GPU / AI deployment | Operational | 1333 | — | Phase I delivered critical IT load, in service at quarter end; 200 MW gross is a different measure. | — | 2026-11-11 |
| CoreWeave Hillsboro | CoreWeave | Hillsboro | United States | GPU / AI deployment | Announced | — | 91 | Announced allocation at each facility in the 2023 agreement. | — | 2026-11-11 |
| CoreWeave Kenilworth NEST 11 | CoreWeave | Kenilworth | United States | GPU / AI deployment | Under construction | — | — | — | — | 2026-11-11 |
| CoreWeave Lancaster | CoreWeave | Lancaster | United States | GPU / AI deployment | Announced | — | 1001 | Initial facility plan; potential expansion to 300 MW | — | 2026-11-11 |
| CoreWeave London Docklands | CoreWeave | London | United Kingdom | GPU / AI deployment | Operational | — | — | — | NVIDIA H200 | 2026-11-11 |
| CoreWeave Marble | CoreWeave | Marble | United States | GPU / AI deployment | Operational | 651 | 651 | About 65 MW billing of 65 MW leased critical IT (Core Scientific, 5 Nov 2026). | — | 2026-11-11 |
| CoreWeave Muskogee | CoreWeave | Muskogee | United States | GPU / AI deployment | Operational | 701 | 701 | About 70 MW billing of 70 MW leased critical IT (Core Scientific, 5 Nov 2026). | — | 2026-11-11 |
| CoreWeave Plano | CoreWeave | Plano | United States | GPU / AI deployment | Announced | — | — | — | — | 2026-11-11 |
| CoreWeave Stockholm 4 South | CoreWeave | Stockholm | Sweden | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| CoreWeave Stockholm second campus | CoreWeave | Stockholm | Sweden | GPU / AI deployment | Announced | — | — | — | — | 2026-11-11 |
| Crusoe Abilene Microsoft | Crusoe | Abilene | United States | GPU / AI deployment | Under construction | — | 9001 | New dedicated campus with on-site power; not operating IT load. | — | 2026-11-11 |
| Crusoe Abilene Stargate | Crusoe | Abilene | United States | GPU / AI deployment | Operational | — | — | — | NVIDIA GB200 | 2026-11-11 |
| Crusoe Houston | Crusoe | Houston | United States | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| Crusoe Keflavik ICE02 | Crusoe | Keflavik | Iceland | GPU / AI deployment | Operational | — | — | — | NVIDIA DGX GB200 NVL72 / Blackwell (expansion) | 2026-11-11 |
| Crusoe Sparks | Crusoe | Sparks | United States | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| Crusoe Tørdal DRA01 | Crusoe | Tørdal | Norway | GPU / AI deployment | Operational | 121 | 401 | Live IT capacity at DRA01. | NVIDIA Vera Rubin NVL72 (DRA02 expansion planned for late 2027) | 2026-11-11 |
| Crusoe Wyoming former project | Crusoe | — | United States | Historical project | Withdrawn / transferred | — | — | — | — | 2026-11-11 |
| Firmus / SMC Loyang | Firmus / SMC | Singapore | Singapore | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| Firmus / SMC Media Hub | Firmus / SMC | Singapore | Singapore | GPU / AI deployment | Status unconfirmed | — | 31 | Media Hub GPU AI factory design capacity. | — | 2026-11-11 |
| Firmus / SMC Southgate Melbourne | Firmus / SMC | Melbourne | Australia | GPU / AI deployment | Operational | 424 | — | About 42 MW deployed when the Southgate partnership with CDC ended (W.Media, 6 Oct 2026). | NVIDIA GB300 NVL72 | 2026-11-11 |
| Firmus Wesley Vale | Firmus / SMC | Wesley Vale | Australia | GPU / AI deployment | Status unconfirmed | — | — | — | — | 2026-11-11 |
| Fluidstack Barber Lake | Fluidstack | Colorado City | United States | GPU / AI deployment | Under construction | — | 1681 | Contracted critical IT load; delivery targeted September 2026 in the filing. | — | 2026-11-11 |
| Fluidstack Bosquel former proposal | Fluidstack | Bosquel | France | Historical project | Withdrawn / transferred | — | — | — | — | 2026-11-11 |
| Fluidstack Lake Mariner | Fluidstack | Barker | United States | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| G42 Abu Dhabi Stargate | G42 | Abu Dhabi | United Arab Emirates | GPU / AI deployment | Under construction | — | 1,0001 | Stargate UAE cluster under construction; distinct from the wider 5 GW technology campus. | — | 2026-11-11 |
| Hyperstack / NexGen Montreal | Hyperstack / NexGen | Montreal | Canada | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| Hyperstack / NexGen NORWAY-1 | Hyperstack / NexGen | — | Norway | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| Hyperstack / NexGen US-1 | Hyperstack / NexGen | — | United States | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| IREN Canal Flats | IREN | Canal Flats | Canada | AI use unconfirmed | Status unconfirmed | — | 301 | Existing campus power; GPU-specific portion undisclosed. | — | 2026-11-11 |
| IREN Childress Horizon 1 | IREN | Childress | United States | GPU / AI deployment | Operational | 501 | — | Horizon 1 critical IT load; excludes Horizon 2–4 and other campus uses | NVIDIA GB300 NVL72 | 2026-11-11 |
| IREN Mackenzie | IREN | Mackenzie | Canada | GPU / AI deployment | Operational | 801 | — | Approximately 80 MW total campus capacity as of 2025-12-31 | — | 2026-11-11 |
| IREN Prince George | IREN | Prince George | Canada | GPU / AI deployment | Operational | 501 | — | Approximately 50 MW total campus capacity as of 2025-12-31 | — | 2026-11-11 |
| IREN Sweetwater | IREN | Sweetwater | United States | GPU / AI deployment | Under construction | — | 2,0001 | Campus power under construction, not GPU IT load. | — | 2026-11-11 |
| Lambda Atlanta ATL02 | Lambda | Atlanta | United States | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| Lambda Chicago AI factory | Lambda | Chicago | United States | GPU / AI deployment | Under construction | — | 231 | Planned single-tenant data center capacity; IT/gross split unspecified | — | 2026-11-11 |
| Lambda Kansas City AI factory | Lambda | Kansas City | United States | GPU / AI deployment | Status unconfirmed | — | 241 | Announced launch capacity; not verified operating capacity | NVIDIA Blackwell Ultra | 2026-11-11 |
| Lambda Mountain View MV1 | Lambda | Mountain View | United States | GPU / AI deployment | Operational | — | — | — | NVIDIA GB300 NVL72 | 2026-11-11 |
| Lambda Plano DFW-04 | Lambda | Plano | United States | GPU / AI deployment | Under construction | — | — | — | — | 2026-11-11 |
| Lambda Vernon LAX01 | Lambda | Vernon | United States | GPU / AI deployment | Operational | 211 | — | Initial Lambda lease allocation; not the full 33 MW facility | NVIDIA Blackwell | 2026-11-11 |
| Nebius Independence | Nebius | Independence | United States | GPU / AI deployment | Under construction | — | 1,2001 | Up to 1.2 GW total campus ambition; not live power | — | 2026-11-11 |
| Nebius Kansas City | Nebius | Kansas City | United States | GPU / AI deployment | Operational | — | 403 | Initial 5 MW, expandable to 40 MW (Nebius, Nov 2024). | NVIDIA H200; NVIDIA B200 | 2026-11-11 |
| Nebius Keflavik Verne | Nebius | Keflavik | Iceland | GPU / AI deployment | Operational | 101 | — | Nebius cluster implementation; not Verne campus total. | — | 2026-11-11 |
| Nebius Lappeenranta | Nebius | Lappeenranta | Finland | GPU / AI deployment | Under construction | — | 3101 | Up to 310 MW at full deployment; IT/gross not stated | — | 2026-11-11 |
| Nebius London | Nebius | London | United Kingdom | GPU / AI deployment | Announced | — | — | — | NVIDIA Blackwell Ultra (contracted deployment) | 2026-11-11 |
| Nebius Madrid | Nebius | Madrid | Spain | GPU / AI deployment | Operational | — | — | — | NVIDIA RTX PRO 6000 | 2026-11-11 |
| Nebius Mäntsälä | Nebius | Mäntsälä | Finland | GPU / AI deployment | Operational | 751 | — | Expanded data center capacity; IT/gross distinction not stated | NVIDIA GB300 NVL72 | 2026-11-11 |
| Nebius near Lille | Nebius | Lille | France | GPU / AI deployment | Under construction | — | 2401 | 240 MW when fully deployed; IT/gross not stated | — | 2026-11-11 |
| Nebius Paris | Nebius | Paris | France | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| Nebius Vineland | Nebius | Vineland | United States | GPU / AI deployment | Operational | — | 3002 | Up to 300 MW phased design capacity (Nebius, March 2025). | — | 2026-11-11 |
| Nebius Woodbury | Nebius | Woodbury | United States | GPU / AI deployment | Operational | — | — | — | NVIDIA B300 NVLink | 2026-11-11 |
| Nscale Blönduós | Nscale | Blönduós | Iceland | GPU / AI deployment | Status unconfirmed | — | — | — | — | 2026-11-11 |
| Nscale Cobalt Park Stargate UK | Nscale | Cobalt Park | United Kingdom | GPU / AI deployment | Announced | — | — | — | — | 2026-11-11 |
| Nscale Glomfjord | Nscale | Glomfjord | Norway | GPU / AI deployment | Operational | 301 | — | Operator-stated operational capacity; expandable to 60 MW; IT/gross split unspecified | — | 2026-11-11 |
| Nscale Hayes | Nscale | Hayes | United Kingdom | GPU / AI deployment | Status unconfirmed | — | — | — | — | 2026-11-11 |
| Nscale Keflavik / Verne | Nscale | Keflavik | Iceland | GPU / AI deployment | Operational | 7.54 | 151 | 7.5 MW deployed in phase 1 of a 15 MW Nscale deployment. | NVIDIA GB300 (Blackwell Ultra) | 2026-11-11 |
| Nscale Loughton | Nscale | Loughton | United Kingdom | GPU / AI deployment | Status unconfirmed | — | 901 | Maximum scalable power allocation; not delivered IT load. | — | 2026-11-11 |
| Nscale Madison | Nscale | Madison | United States | GPU / AI deployment | Announced | — | 401 | Capacity expected online in 2026; 99 MW secured power is a separate scope. | — | 2026-11-11 |
| Nscale Monarch | Nscale | — | United States | GPU / AI deployment | Announced | — | 1,3501 | Phase 1 engineered AI capacity; target early 2028. | — | 2026-11-11 |
| Nscale Narvik / Kvandal | Nscale | Narvik | Norway | GPU / AI deployment | Under construction | — | 2301 | Planned campus capacity; not verified live IT power | NVIDIA Rubin (planned) | 2026-11-11 |
| Nscale Oslo | Nscale | Oslo | Norway | GPU / AI deployment | Status unconfirmed | — | — | — | — | 2026-11-11 |
| Nscale Sines / Start Campus | Nscale | Sines | Portugal | GPU / AI deployment | Operational | — | — | — | NVIDIA Blackwell Ultra | 2026-11-11 |
| Nscale Stavanger | Nscale | Stavanger | Norway | GPU / AI deployment | Status unconfirmed | — | — | — | — | 2026-11-11 |
| Nscale Ward County | Nscale | Ward County | United States | GPU / AI deployment | Operational | — | 2001 | Approximate IT load supported by financed deployment; not verified live draw | NVIDIA GB300 and Vera Rubin VR200 (financed deployment) | 2026-11-11 |
| Sesterce Bosquel | Sesterce | Bosquel | France | GPU / AI deployment | Announced | — | 1,0001 | Planned campus or first-phase capacity; no commissioning confirmation. | — | 2026-11-11 |
| Sesterce Jämsä | Sesterce | Jämsä | Finland | GPU / AI deployment | Announced | — | 2001 | Planned campus or first-phase capacity; no commissioning confirmation. | — | 2026-11-11 |
| Sesterce Valence | Sesterce | Valence | France | GPU / AI deployment | Announced | — | 401 | Planned campus or first-phase capacity; no commissioning confirmation. | — | 2026-11-11 |
| Together AI Maryland 5C | Together AI | — | United States | GPU / AI deployment | Operational | — | — | — | NVIDIA B200 | 2026-11-11 |
| Together AI Memphis 5C | Together AI | Memphis | United States | GPU / AI deployment | Announced | — | — | — | NVIDIA GB200 and GB300 (planned) | 2026-11-11 |
| Voltage Park Allen | Voltage Park | Allen | United States | GPU / AI deployment | Operational | — | — | — | NVIDIA H100 80 GB SXM5 | 2026-11-11 |
| Voltage Park Quincy | Voltage Park | Quincy | United States | GPU / AI deployment | Operational | — | — | — | NVIDIA H100 80 GB SXM5 | 2026-11-11 |
| Vultr Amsterdam | Vultr | Amsterdam | Netherlands | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Atlanta | Vultr | Atlanta | United States | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Bangalore | Vultr | Bangalore | India | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Chennai | Vultr | Chennai | India | GPU / AI deployment | Operational | 201 | — | Opening announcement supercluster power; scope not specified as IT or facility. | AMD Instinct MI355X | 2026-11-11 |
| Vultr Chicago | Vultr | Chicago | United States | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Dallas | Vultr | Dallas | United States | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Delhi NCR | Vultr | Delhi NCR | India | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Frankfurt | Vultr | Frankfurt | Germany | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Honolulu | Vultr | Honolulu | United States | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Johannesburg | Vultr | Johannesburg | South Africa | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr London | Vultr | London | United Kingdom | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Los Angeles | Vultr | Los Angeles | United States | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Madrid | Vultr | Madrid | Spain | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Manchester | Vultr | Manchester | United Kingdom | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Melbourne | Vultr | Melbourne | Australia | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Mexico City | Vultr | Mexico City | Mexico | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Miami | Vultr | Miami | United States | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Milan | Vultr | Milan | Italy | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Mumbai | Vultr | Mumbai | India | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Osaka | Vultr | Osaka | Japan | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Paris | Vultr | Paris | France | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Piscataway | Vultr | Piscataway | United States | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr San Jose | Vultr | San Jose | United States | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Santiago | Vultr | Santiago | Chile | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr São Paulo | Vultr | São Paulo | Brazil | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Seattle | Vultr | Seattle | United States | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Seoul | Vultr | Seoul | South Korea | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Singapore | Vultr | Singapore | Singapore | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Springfield | Vultr | Springfield | United States | GPU / AI deployment | Announced | — | 501 | Announced additional supercluster deployment; no commissioning confirmation in this release. | AMD Instinct MI355X | 2026-11-11 |
| Vultr Stockholm | Vultr | Stockholm | Sweden | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Sydney | Vultr | Sydney | Australia | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Tel Aviv | Vultr | Tel Aviv | Israel | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Tokyo | Vultr | Tokyo | Japan | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Toronto | Vultr | Toronto | Canada | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Vultr Warsaw | Vultr | Warsaw | Poland | General cloud footprint | Operational | — | — | — | — | 2026-11-11 |
| Yotta D1 Shakti Cloud | Yotta | Greater Noida | India | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
| Yotta D2 Blackwell | Yotta | Greater Noida | India | GPU / AI deployment | Announced | — | 601 | D2 data center capacity; campus expansion is separate. | NVIDIA Blackwell Ultra (planned) | 2026-11-11 |
| Yotta NM1 Shakti Cloud | Yotta | Navi Mumbai | India | GPU / AI deployment | Operational | — | — | — | — | 2026-11-11 |
Map of neocloud locations
Positioning methodPick a numbered pin to jump to its sites, then open any site file for power, GPUs and sources.
World
115 mapped sites- Boyce / Dallas / Denton / Austin / Douglasville / Afton / Muskogee / Plano / Abilene / Houston / Colorado City / Childress / Sweetwater / Atlanta / Memphis / Allen Applied Digital Delta Forge 1Core42 DallasCoreWeave at DentonCoreWeave AustinCoreWeave DouglasvilleCoreWeave HeliosCoreWeave MuskogeeCoreWeave PlanoCrusoe Abilene MicrosoftCrusoe Abilene StargateCrusoe HoustonFluidstack Barber LakeIREN Childress Horizon 1IREN SweetwaterLambda Atlanta ATL02Lambda Plano DFW-04Together AI Memphis 5CVoltage Park AllenVultr AtlantaVultr Dallas
- Harwood / Minneapolis / Ellendale / Woodbury Applied Digital Polaris Forge 2Core42 MinneapolisCoreWeave Ellendale Polaris Forge 1Nebius Woodbury
- Barker / Kenilworth / Lancaster / Montreal / Vineland / Piscataway / Springfield / Toronto Core42 Lake MarinerCoreWeave Kenilworth NEST 11CoreWeave LancasterFluidstack Lake MarinerHyperstack / NexGen MontrealNebius VinelandVultr PiscatawayVultr SpringfieldVultr Toronto
- Stockton / Sunnyvale / Sparks / Mountain View / Vernon / Los Angeles / San Jose Core42 StocktonCore42 SunnyvaleCrusoe SparksLambda Mountain View MV1Lambda Vernon LAX01Vultr Los AngelesVultr San Jose
- Vennesla / Copenhagen / Falun / Oslo / Stavanger CoreWeave at Bulk N01CoreWeave Copenhagen DEN01CoreWeave FalunNscale OsloNscale Stavanger
- Barcelona / Madrid / Valence CoreWeave Barcelona Zona FrancaNebius MadridSesterce ValenceVultr Madrid
- Crawley / London / Bosquel / Lille / Paris / Hayes / Loughton / Amsterdam / Frankfurt / Manchester CoreWeave CrawleyCoreWeave London DocklandsFluidstack Bosquel former proposalNebius LondonNebius near LilleNebius ParisNscale HayesNscale LoughtonSesterce BosquelVultr AmsterdamVultr FrankfurtVultr LondonVultr ManchesterVultr Paris
- Dalton / Marble / Madison CoreWeave DaltonCoreWeave MarbleNscale Madison
- Hillsboro / Quincy / Seattle CoreWeave HillsboroVoltage Park QuincyVultr Seattle
- Stockholm / Mäntsälä / Jämsä CoreWeave Stockholm 4 SouthCoreWeave Stockholm second campusNebius MäntsäläSesterce JämsäVultr Stockholm
- Keflavik / Blönduós Crusoe Keflavik ICE02Nebius Keflavik VerneNscale BlönduósNscale Keflavik / Verne
- Singapore Firmus / SMC LoyangFirmus / SMC Media HubVultr Singapore
- Melbourne / Wesley Vale / Sydney Firmus / SMC Southgate MelbourneFirmus Wesley ValeVultr MelbourneVultr Sydney
- Abu Dhabi G42 Abu Dhabi Stargate
- Canal Flats / Prince George IREN Canal FlatsIREN Prince George
- Mackenzie IREN Mackenzie
- Chicago / Kansas City / Independence Lambda Chicago AI factoryLambda Kansas City AI factoryNebius IndependenceNebius Kansas CityVultr Chicago
- Lappeenranta Nebius Lappeenranta
- Glomfjord / Narvik Nscale GlomfjordNscale Narvik / Kvandal
- Sines Nscale Sines / Start Campus
- Chennai Vultr Chennai
- Honolulu Vultr Honolulu
- Johannesburg Vultr Johannesburg
- Mexico City Vultr Mexico City
- Miami Vultr Miami
- Milan Vultr Milan
- Mumbai / Navi Mumbai Vultr MumbaiYotta NM1 Shakti Cloud
- Osaka / Tokyo Vultr OsakaVultr Tokyo
- Santiago Vultr Santiago
- São Paulo Vultr São Paulo
- Seoul Vultr Seoul
- Tel Aviv Vultr Tel Aviv
- Warsaw Vultr Warsaw
- Greater Noida Yotta D1 Shakti CloudYotta D2 Blackwell
North America
55 mapped sites- Boyce Applied Digital Delta Forge 1
- Harwood / Ellendale Applied Digital Polaris Forge 2CoreWeave Ellendale Polaris Forge 1
- Dallas / Denton / Plano / Allen Core42 DallasCoreWeave at DentonCoreWeave PlanoLambda Plano DFW-04Voltage Park AllenVultr Dallas
- Barker / Toronto Core42 Lake MarinerFluidstack Lake MarinerVultr Toronto
- Minneapolis / Woodbury Core42 MinneapolisNebius Woodbury
- Stockton / Sunnyvale / Sparks / Mountain View / San Jose Core42 StocktonCore42 SunnyvaleCrusoe SparksLambda Mountain View MV1Vultr San Jose
- Austin CoreWeave Austin
- Dalton / Douglasville / Marble / Atlanta CoreWeave DaltonCoreWeave DouglasvilleCoreWeave MarbleLambda Atlanta ATL02Vultr Atlanta
- Afton / Abilene / Colorado City / Childress / Sweetwater CoreWeave HeliosCrusoe Abilene MicrosoftCrusoe Abilene StargateFluidstack Barber LakeIREN Childress Horizon 1IREN Sweetwater
- Hillsboro / Seattle CoreWeave HillsboroVultr Seattle
- Kenilworth / Lancaster / Vineland / Piscataway CoreWeave Kenilworth NEST 11CoreWeave LancasterNebius VinelandVultr Piscataway
- Muskogee CoreWeave Muskogee
- Houston Crusoe Houston
- Montreal Hyperstack / NexGen Montreal
- Canal Flats IREN Canal Flats
- Mackenzie / Prince George IREN MackenzieIREN Prince George
- Chicago Lambda Chicago AI factoryVultr Chicago
- Kansas City / Independence Lambda Kansas City AI factoryNebius IndependenceNebius Kansas City
- Vernon / Los Angeles Lambda Vernon LAX01Vultr Los Angeles
- Madison Nscale Madison
- Memphis Together AI Memphis 5C
- Quincy Voltage Park Quincy
- Miami Vultr Miami
- Springfield Vultr Springfield
Europe
38 mapped sites- Vennesla CoreWeave at Bulk N01
- Barcelona CoreWeave Barcelona Zona Franca
- Copenhagen CoreWeave Copenhagen DEN01
- Crawley / London / Hayes / Loughton CoreWeave CrawleyCoreWeave London DocklandsNebius LondonNscale HayesNscale LoughtonVultr London
- Falun CoreWeave Falun
- Stockholm CoreWeave Stockholm 4 SouthCoreWeave Stockholm second campusVultr Stockholm
- Keflavik Crusoe Keflavik ICE02Nebius Keflavik VerneNscale Keflavik / Verne
- Bosquel / Lille / Paris Fluidstack Bosquel former proposalNebius near LilleNebius ParisSesterce BosquelVultr Paris
- Lappeenranta Nebius Lappeenranta
- Madrid Nebius MadridVultr Madrid
- Mäntsälä / Jämsä Nebius MäntsäläSesterce Jämsä
- Blönduós Nscale Blönduós
- Glomfjord Nscale Glomfjord
- Narvik Nscale Narvik / Kvandal
- Oslo Nscale Oslo
- Sines Nscale Sines / Start Campus
- Stavanger Nscale Stavanger
- Valence Sesterce Valence
- Amsterdam Vultr Amsterdam
- Frankfurt Vultr Frankfurt
- Manchester Vultr Manchester
- Milan Vultr Milan
- Warsaw Vultr Warsaw
In the table, without a verified locality pin: CoreWeave Arasur, Crusoe Tørdal DRA01, Crusoe Wyoming former project, Hyperstack / NexGen NORWAY-1, Hyperstack / NexGen US-1, Nscale Cobalt Park Stargate UK, Nscale Monarch, Nscale Ward County, Together AI Maryland 5C, Vultr Bangalore, Vultr Delhi NCR.
Locality centers, not facility addresses. Nearby localities share a numbered group; pin sizes do not encode capacity. The world map covers all verified locality pins; regional plates enlarge North America and Europe. Maps are built into the HTML and remain independent of table filters. Land: Natural Earth, public domain. Positions: Open-Meteo / GeoNames, locality source links in each record.
Start with your deployment constraint
- North America61 sites →
US and Canadian deployments, from dedicated GPU halls to new campus plans.
- Europe42 sites →
Compare Nordic, British and continental European physical deployments.
- Operating deployments84 sites →
Locations with a public operational statement. Expansion capacity may still be planned.
- Construction & plans40 sites →
Announced facilities, construction projects and unresolved delivery status.
Neocloud sites by country
Count source: dataset- Australia4 sites →
- Brazil1 sites →
- Canada5 sites →
- Chile1 sites →
- Denmark1 sites →
- Finland3 sites →
- France6 sites →
- Germany1 sites →
- Iceland4 sites →
- India7 sites →
- Israel1 sites →
- Italy1 sites →
- Japan2 sites →
- Mexico1 sites →
- Netherlands1 sites →
- Norway7 sites →
- Poland1 sites →
- Portugal1 sites →
- Singapore3 sites →
- South Africa1 sites →
- South Korea1 sites →
- Spain4 sites →
- Sweden4 sites →
- United Arab Emirates1 sites →
- United Kingdom8 sites →
- United States56 sites →
Neocloud sites by operator
Coverage limits- Applied Digital2 sites →
- Core425 sites →
- CoreWeave21 sites →
- Crusoe7 sites →
- Firmus / SMC4 sites →
- Fluidstack3 sites →
- G421 sites →
- Hyperstack / NexGen3 sites →
- IREN5 sites →
- Lambda6 sites →
- Nebius11 sites →
- Nscale13 sites →
- Sesterce3 sites →
- Together AI2 sites →
- Voltage Park2 sites →
- Vultr35 sites →
- Yotta3 sites →
Field notes / public evidence
A location is only useful with its scope.
We read operator pages, hosting partners’ specifications and filings. Each record links individual fields to their evidence and separates the date of the statement from the date we checked it.
A site turns Operational when the operator, its landlord or a filing reports it live. GPU counts and MW are tied to the site they describe, never spread from a company-wide total.
Questions about neocloud locations
What is a neocloud location?
A physical data center where a specialist GPU cloud such as CoreWeave, Nebius, Lambda, Nscale or Crusoe runs or is building capacity. Each record names the city, the host or partner, power and GPU platform where published, and the source behind every value.
What do delivered MW and planned MW mean?
Delivered MW is capacity the operator, its landlord or a filing reports as live or billing. Planned MW is the contracted or announced total and can include delivered phases. The power basis beside each value states its exact scope, so you compare like with like.
How precise are the map pins?
Each pin sits on the GeoNames center of the named city, so it shows the metro rather than the building. Nearby cities share one numbered pin, and the list under each map names every site in the group.
Can I buy GPU capacity at these sites?
Yes, at most of them. Operators sell on-demand, reserved and dedicated capacity from these locations, and some halls are built for a single customer such as Microsoft or OpenAI. Use Request a quote on any site page to reach operators with capacity in that region.
Why are some cells empty?
A cell fills only when a public source gives a figure for that specific site. Provider-wide fleets are never split across locations, so every number on the map traces back to a source for that place.